Some things have more uncertainty than others (compared to what the number would be used for) but we have created techniques to deal with that in practise too (hedging, optionality.)
Maybe you have a personal threshold for a level of uncertainty below which things count as measurable and above which they don't, but you're imbuing the word "measure" with subjective characteristics at that part -- not to be confused with its objective meaning.
Estimation is on the other hand uses measurements as input and has no clear cut path forward. (you must assume some model)
"Chemistry is just applied physics" ("measurement is just estimating via proxy") might be technically correct, but distinction is still very real and usefull.
I'd argue probability is subjective and estimation is mostly unambiguous, therefore the estimation uncertainty is mostly known, thus estimation is a form of measurement.
I can absolutely see how, if you think probability is frequentist/objective or you think there's generally non-uncertain ambiguity in the estimation task, you would disagree.
That is a very generous take. It can be estimated or guessed at via proxies, but something like that is not really measurable. It's like measuring the loss of a key employee leaving.